During the year ending June 30, 1904, the numbers reported as killed and injured were 10,046 and 84,155 respectively. This is an average of one person killed every 52 minutes and one person injured every 6 minutes. While this statement looks very bad, the railroads are not altogether responsible. Of the total 10,046, only 441 (a little over 4%) were passengers; 3632 (about 36%) were employees; and the remainder, 5973, were "other persons," of whom 5105 (over 50% of the total) were "trespassers," which includes suicides. Therefore the railroads cannot be held responsible for over one-half of those deaths. Of the 84,155 persons injured, the number of passengers was 9111, or about 11%, and the number of employees 67,067, about 80%. About one-sixth of the deaths, and about one-eighth of the injuries, of passengers were caused by "jumping on or off trains, locomotives, or cars." In nearly all of such cases the passengers were alone responsible.

In the face of such a death-list, it is hard to realize the very small probability of an injury or death occurring to any passenger during any one ride. The "number of passengers carried one mile" for 1904 equaled 21,923,213,536. Dividing this by 441, the number of deaths of passengers, we have over 49,700,000, the number of passenger-miles for each death. Again, dividing 21,923,213,536 by 9111, the number of passengers injured, we have over 2,400,000 as the number of passenger-miles for each injury. The practical meaning of such figures is that, in the language of the theory of probabilities, the "chances are even" that if a passenger were to ride continuously at the rate of 40 miles per hour, or 350,640 miles per year, he would ride for 142 years before being killed, or about 7 years before being injured. But, as a matter of fact, no one uses the railway for a distance of 350,000 miles per year, or even any large proportion of it. A better way of considering the probabilities would be to say that, since there were 49,700,000 passenger-miles for each death, a trip of say 100 miles would involve a risk of one in 497,000 that it might result in death, assuming that the operation of the railroad during that trip was neither more nor less hazardous than the average railroad operation. A similar figure could be computed to determine the probability that any given trip might result in an injury to any given passenger. It should not be forgotten that during the year 1904 a far greater number of passengers were killed than during any year for the previous 16 years. In 1895 the number of passengers killed was only 170, which was less than 40% of the figure for 1904. Although the amount of railway business was far greater in 1904 than in 1895, and we might expect some increase in the fatalities, yet the probabilities were far less in 1895 than in 1904.

These figures for 1904 have been retained in the second edition, although later figures are obtainable, because, although the probability of death or injury to a passenger is always so small as to be practically negligible, as shown in the last paragraph, the figures for the year 1909, the latest which are available, show even less probability. Although the number of passengers carried one mile increased about one-fourth in five years, the total number of passengers killed in 1909 was less than 60% of those killed in 1904. While this may be a mere fluctuation of figures, a comparison of the figures for many years past show that those for 1904 are abnormally high.

There has been much discussion of the subject of accidents as an element in railroad economics. Some economists have endeavored to place a financial value on accidents and to determine how much money could profitably be spent to avoid the danger of accidents. While no one will deny the justification of spending a reasonable amount of money to avoid the danger of an accident due to some specific cause, the question always remains, How much actual lessening of danger will be accomplished by any reasonable or practicable expenditure of money? Certain classes of improvements are demonstrably justifiable, as, for example, the elimination of grade highway-crossings, especially on a road of heavy traffic. A more doubtful question occurs when it is proposed to reduce some sharp curvature when passing through a mountainous region where the view of the track is obstructed for any great distance. In such a case the practical question is not the lessening of danger by the elimination of all curvature, which would probably be financially, if not physically, impossible, but the lessening of danger by the use of less curvature rather than more. It may usually be shown that the lessening of the probability of accidents, due to such reductions of curvature as are practicable, is so small that such a consideration alone will not justify the expenditure of any appreciable amount of money to accomplish the result.

6. Use Of Statistical Averages

The student should be cautioned against an improper use of the statistical averages given in this chapter and in Chapter VI. They should be considered somewhat in the same light as a composite photograph of a group of men. The composite photograph cannot be considered as a correct photograph of anyone, unless he happens to have the average features. The above averages regarding wealth per capita, railway capital per capita, and the average payment to railroads per capita should not be assumed to have any application to any particular case. For example, the student should not consider that the average annual payment as given for 1904 - \$24 per capita - would represent the earnings of any proposed railroad from its tributary population, unless there is good reason to believe that the particular territory in question is a fair average sample of the whole country. And even in this case the determination of the " tributary population" is not easy. Such a figure has its value to give the student a rough idea of railroad earnings and railroad operation, but he should know that such figures cannot be depended on, except as a rough check on computations which have been more carefully made from local considerations.